Jump to content

Leaderboard

Popular Content

Showing content with the highest reputation on 11/11/25 in Posts

  1. Just sharing some good news — my update went smoothly, and all apps, VMs, and backups are running normally under Fangtooth.
    1 point
  2. Thanks for reporting these issues. We are taking this feedback to heart and will work to address.
    1 point
  3. Hello! Are there plans for HexOS to allow users to use (mirrored) SSDs for ZFS' tiered storage optimizations? The main three uses I'm thinking of: Enable high-endurance mirrored SSDs to accept synchronous writes (SLOG) Enable mirrored SSDs to store all file metadata (special vdev) Enable mirrored SSDs to store all small files (special vdev) While I'm sure popping into TrueNAS (for those with that experience) would work, I think implementing these directly in HexOS would be quite beneficial even for consumers with many small files and / or macOS devices. While not all performance-based vdevs are helpful in most scenarios, these would be great, especially the special vdev for small files, e.g., use the ultra-fast random access of SSDs for video editing project files or documents (special vdev) and especially the SLOG vdev for macOS SMB writes (always sync writes). Of course, in some far and away future, HexOS would identify when these vdevs would significantly improve performance, but manually turning them on & testing them out would also be neat through HexOS. I'm intentionally ignoring L2ARC, which requires much more careful hardware selection & tuning to get any benefit.
    1 point
  4. hexos will not show the drives as used as support for special vdev in the hexos interface was never added. once a special vdev is added to a pool, their use is automatic. You don't have to worry about hexos seeing or not seeing it correctly.
    1 point
  5. Thank you for your reply, @Mobius. That does make sense. To your final note, hopefully by stable, if we manually create these vdevs in TrueNAS, they can play nicely with HexOS. — I agree: I believe folks recommend the special vdev to have the same redundancy as the data vdevs; if it were to be added, hopefully that recommendation could be prominent. I've heard mixed things about performance: for many-file directories, it can feel snappier, which is what I hope. SLOG: ah, I had another angle. My main interest are a couple of wired Macs on our network as Macs write to SMB synchronously (straight to HDDs), without fast RAM-first writes like Windows machines. But with an SLOG, Macs write to this faster SLOG vdev, allowing the transfer to complete faster. So with an SLOG, Macs have virtually identical write performance as Windows. But without the SLOG, Macs write a little slower. We definitely use a UPS; it's a solid investment for data integrity as you get PLP for the whole server! A great recommendation.
    1 point
  6. i'm not gonna go as far as saying never gonna happen but i'd say its on the furthest back burner if at all. special vdev is a scary one to implement because if it goes down, itll take the whole pool down with it on top of not being very useful for most homelab users slog also doesn't make the most sense for most users as to my best understanding it only really protects you in case of power loss and at that point you might be better served just getting a ups. Personally i do run a triple mirror special vdev on my storage pool and i think it causes some stability issues with hexos.
    1 point
×
×
  • Create New...